Selective Backlight Correction System for Image Visibility Enhancement

Abstract

The present invention is a backlight correction system for improving the visibility of an input image captured in a backlight environment, comprising: a region segmentation unit that detects a problem area where illuminance imbalance occurs by analyzing the luminance distribution at the pixel level from an input image frame; a backlight correction unit that, corresponding to the problem area detected by the region segmentation unit, separates the illumination component and the reflection component based on a brightness channel and corrects the illumination component while preserving the reflection component to mitigate contrast distortion caused by backlight or saturation; a fusion unit that generates a final image by naturally combining the problem area corrected by the backlight correction unit and the non-problem area of ​​the original image through boundary blending; and a control unit that is respectively connected to the region segmentation unit, the backlight correction unit, and the fusion unit to control the operation of the region segmentation unit, the backlight correction unit, and the fusion unit, wherein the backlight correction unit selectively performs correction only on the problem area under the control of the control unit and maintains the original image information of the non-problem area.

Description

Technology Field

[0001] The present invention relates to the field of image processing technology for improving the visibility of images captured in backlit or non-uniform illumination environments, and to a selective backlight correction system for improving image visibility that mitigates contrast and color distortion by analyzing the luminance distribution of an input image and selectively correcting problem areas. Background Technology

[0003] Video surveillance and analysis technologies have played a pivotal role in various fields, including urban safety, traffic management, industrial facility monitoring, and disaster response. In particular, CCTV-based video surveillance systems have established themselves as essential infrastructure for crime prevention and post-crime analysis. While such systems must be able to reliably acquire and analyze footage under diverse environmental conditions regardless of day or night, actual shooting environments are significantly affected by numerous external factors, such as the position of the sun, seasonal changes, weather conditions, and the placement of artificial lighting. Among these, backlighting is a major factor causing image quality degradation; when a strong light source is present in the background of a subject, the subject is frequently rendered excessively dark or perceived only as a silhouette, resulting in the loss of information regarding the object's shape, color, and detailed contours. This phenomenon is very common in actual CCTV installation environments, such as entrances, windows, tunnel entrances, parking lot access points, road intersections, and outdoor surveillance zones, and significantly reduces the accuracy of video-based situational awareness and object identification.

[0004] Conventional image processing techniques have primarily utilized methods that uniformly adjust the brightness or contrast of the entire image. For instance, traditional contrast enhancement techniques such as gamma correction, histogram stretching, and histogram equalization have been applied to various devices due to their relatively simple implementation and low computational burden. However, because these techniques process the entire image using the same standard, they have limitations: attempting to improve dark areas caused by backlighting can lead to saturation by excessively amplifying already sufficiently bright areas, while attempting to suppress bright areas results in the loss of detail in normal areas or within the subject. Particularly in backlit environments, where dark and very bright areas often coexist, it is difficult to secure natural and balanced image quality using only global correction methods. Consequently, while image visibility may be partially improved, concerns have been continuously raised regarding the insufficient reliability of information required for actual surveillance and analysis purposes.

[0005] Furthermore, methods to attempt optical improvements in terms of camera hardware have also been proposed. For instance, technologies have been commercialized that incorporate Auto Exposure (AE), Automatic White Balance (AWB), and Wide Dynamic Range (WDR) functions into the camera itself to acquire relatively balanced images even in scenes with significant contrast differences. However, this hardware-based approach is subject to constraints such as sensor performance, manufacturing costs, and power consumption, and has limitations in that it is difficult to flexibly apply it via software to existing CCTV systems. Additionally, while technologies like WDR may be effective under specific conditions, they can still result in loss of subject information or artificial image distortion in extreme backlit environments or complex lighting conditions. Consequently, there is a problem where image quality is inconsistent and results vary significantly depending on the scene.

[0006] Recently, alongside the advancement of artificial intelligence and deep learning technologies, learning-based approaches are being actively researched in the field of image enhancement. Convolutional Neural Networks (CNNs) or Transformer-based models have been proposed to address various issues, such as low-light image enhancement, fog removal, rain and snow removal, and noise reduction; some technologies are also being applied to image enhancement in backlit environments. However, deep learning-based technologies require massive amounts of training data and high computational loads, often necessitating high-performance GPUs or dedicated accelerators. This imposes limitations on real-time processing in edge environments or low-power devices and can lead to processing delays or increased costs in systems that must handle multiple channels simultaneously, such as CCTVs. Furthermore, it has been pointed out that generalization performance may degrade under new lighting conditions or environmental changes not included in the training data, and that it is difficult to ensure the stability and consistency required for surveillance systems because image enhancement results are unpredictable.

[0007] Meanwhile, backlighting issues extend beyond merely detracting from the aesthetics of images; they directly impact the performance of downstream video analysis algorithms, such as object detection, tracking, and behavior recognition. If backlighting obscures object boundaries or distorts color information, the likelihood of false positives or failures in object recognition algorithms increases, potentially leading to security incidents or delays in situational awareness. Consequently, video enhancement technology is recognized as a foundational technology capable of supporting the reliability and accuracy of entire video analysis systems, going beyond simple image quality improvement. In particular, as CCTV footage is increasingly utilized for various purposes—including real-time monitoring, incident analysis, evidence gathering, and AI-based automated surveillance—preserving original video information and achieving natural visual restoration have become increasingly critical.

[0008] Against this backdrop, the demand for image correction technology specialized for backlit environments is continuously increasing. There is a growing emphasis on an approach that minimizes information loss in the original footage by precisely identifying areas where illumination imbalances actually occur and selectively performing correction only on those specific regions, rather than simply processing the entire video in a batch. Furthermore, such technology must be applicable to real-time video streams and capable of flexible integration into various system environments, including existing CCTVs, NVRs, and edge computing devices. Additionally, it is required to provide visually natural results to prevent artificial boundaries or distortions from occurring during the correction process. Ultimately, backlight correction technology needs to evolve in a direction that simultaneously satisfies improved visibility, information preservation, real-time performance, and system compatibility, taking into account the diversity and complexity of video acquisition environments. These demands serve as the technological foundation for complementing the limitations of existing technologies and seeking a new paradigm for image processing. Prior art literature

[0010] Republic of Korea Registered Patent No. 10-1257942 (April 23, 2013) Republic of Korea Published Patent No. 1993-0016636 The problem to be solved

[0011] Accordingly, the present invention has been devised to solve the aforementioned problems, and the objective of the present invention is to provide a selective backlight correction system for improving image visibility, which improves the problems of contrast distortion and detail loss that impede the visibility of a subject in images captured in backlight or uneven illumination environments, thereby enabling clearer recognition of the shape, contour, and color information of an object.

[0012] Another objective of the present invention is to provide a selective backlight correction system for improving image visibility that overcomes the limitations of conventional methods of processing the entire image collectively and performs selective correction only on the areas where problems actually occur, thereby minimizing information loss of the original image and maintaining natural image quality.

[0014] However, the technical problems to be solved by the present invention are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art to which the present invention belongs from the description below. means of solving the problem

[0016] The present invention was created to improve upon the problems of the prior art described above. It is a backlight correction system for improving the visibility of an input image captured in a backlight environment, comprising: a region segmentation unit that detects a problem area where illuminance imbalance occurs by analyzing the luminance distribution at the pixel level from an input image frame; a backlight correction unit that, corresponding to the problem area detected by the region segmentation unit, separates the illumination component and the reflection component based on a brightness channel and corrects the illumination component while preserving the reflection component to mitigate contrast distortion caused by backlight or saturation; a fusion unit that generates a final image by naturally combining the problem area corrected by the backlight correction unit and the non-problem area of ​​the original image through boundary blending; and a control unit that is respectively connected to the region segmentation unit, the backlight correction unit, and the fusion unit to control the operation of the region segmentation unit, the backlight correction unit, and the fusion unit. The backlight correction unit can selectively perform correction only on the problem area under the control of the control unit and maintain the original image information of the non-problem area.

[0018] In addition, in one embodiment, the region segmentation unit may calculate a luminance distribution by performing a histogram-based statistical analysis on the RGB channels of an input image, and generate a luminance map based on pixel brightness based on the luminance distribution.

[0020] In addition, in one embodiment, the region segmentation unit can separate the brightness channel by color space converting the illuminance map, generate a backlight region mask and a saturated region mask respectively according to a preset threshold, and combine them to generate a final problem region mask.

[0022] In addition, in one embodiment, the backlight correction unit approximates the observed value of the input image as the product of the illumination component and the reflection component, and generates an illumination-invariant color map by normalizing the illumination component to a reference value, thereby maintaining the relative color ratio of the subject even in a backlight environment.

[0024] In addition, in one embodiment, the illumination-invariant color map may be selectively applied only to the area where the problem area mask is applied, and controlled so that the original image is maintained in areas other than the problem area.

[0026] In addition, in one embodiment, the backlight correction unit can balance the contrast by performing a contrast restoration operation in a logarithmic space to enhance details in areas that are excessively dark from a set standard, and by suppressing highlights in saturated areas through brightness tone mapping.

[0028] In addition, in one embodiment, the backlight correction unit dynamically adjusts the correction intensity according to at least one of the frame-by-frame average illuminance change of the input image or the area ratio of the problem area, thereby enabling improved visibility even in an environment with lighting changes exceeding a set standard.

[0030] In addition, in one embodiment, the fusion unit may apply at least one of weight-based blending or multi-resolution mixing techniques to reduce artificial boundary phenomena that may occur at the boundary between the corrected area and the uncorrected area.

[0032] In addition, in one embodiment, the backlight correction system is mounted on at least one of a CCTV camera, a network video recorder (NVR), and an edge computing device, and can perform backlight correction in real time on an input video stream. Effects of the invention

[0034] According to one embodiment of the present invention, the brightness and color of areas that have become dark or saturated due to backlighting are effectively restored, thereby improving the overall visibility of the image and significantly improving the accuracy of situational awareness, object identification, and analysis based on CCTV video.

[0035] In addition, according to one embodiment of the present invention, by naturally combining the corrected area and the uncorrected area, artificial boundaries or visual distortions can be suppressed, thereby providing the effect of providing stable and highly reliable image quality even in real-time video surveillance and edge environments.

[0037] However, the effects obtainable from the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0039] The following drawings attached to this specification illustrate preferred embodiments of the present invention and serve to further enhance understanding of the technical concept of the present invention together with the detailed description of the invention provided below; therefore, the present invention should not be interpreted as being limited only to the matters described in such drawings. FIG. 1 is an overall conceptual diagram of a selective backlight correction system for improving image visibility according to an embodiment of the present invention. Figure 2 is a control block diagram of the components of the control unit. FIG. 3 is a diagram illustrating the overall flow of a selective backlight correction system for improving image visibility according to an embodiment of the present invention. Specific details for implementing the invention

[0040] Below, with reference to the attached drawings, embodiments of the present invention are described in detail so that those skilled in the art can easily implement the invention. However, since the description of the present invention is merely an example for structural or functional explanation, the scope of the present invention should not be interpreted as being limited by the embodiments described in the text. That is, since the embodiments are subject to various modifications and may take various forms, the scope of the present invention should be understood to include equivalents capable of realizing the technical concept. Furthermore, the objectives or effects presented in the present invention do not imply that a specific embodiment must include all of them or only such effects; therefore, the scope of the present invention should not be understood as being limited by them.

[0041] The meaning of the terms described in this invention should be understood as follows.

[0042] Terms such as "first" and "second" are intended to distinguish one component from another, and the scope of rights shall not be limited by these terms. For example, the first component may be named the second component, and similarly, the second component may be named the first component. When a component is referred to as being "connected" to another component, it should be understood that it may be directly connected to that other component, or that there may be other components in between. Conversely, when a component is referred to as being "directly connected" to another component, it should be understood that there are no other components in between. Meanwhile, other expressions describing the relationship between components, such as "between" and "exactly between," or "adjacent to" and "directly adjacent to," shall be interpreted in the same manner.

[0043] A singular expression should be understood to include a plural expression unless the context clearly indicates otherwise, and terms such as "include" or "have" are intended to specify the existence of the set-up features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood not to preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0044] Unless otherwise defined, all terms used herein have the same meaning as generally understood by those skilled in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted as having meanings consistent with the context of the relevant technology and should not be interpreted as having an ideal or overly formal meaning unless explicitly defined in this invention.

[0046] FIG. 1 is an overall conceptual diagram of a selective backlight correction system for improving image visibility according to an embodiment of the present invention, FIG. 2 is a block diagram of the component control of a control unit, and FIG. 3 is a diagram illustrating the overall flow of a selective backlight correction system for improving image visibility according to an embodiment of the present invention.

[0047] As illustrated in FIG. 1 and FIG. 3, the present invention is a backlight correction system for improving the visibility of an input image captured in a backlight environment, and may include a region division unit (100), a backlight correction unit (200), a fusion unit (300), and a control unit (400).

[0049] First, regarding the input backlight image, the backlight correction system (10) according to an embodiment of the present invention first performs a pixel-unit illumination distribution analysis on each image frame to detect problem areas where illumination imbalance occurs due to being excessively dark or excessively bright within the image. To this end, a histogram-based statistical analysis is performed for each RGB color channel, and after calculating the distribution of pixel brightness values, an illumination map reflecting brightness values ​​relative to the channel average is generated. This illumination map is used as a criterion for determining where in the image illumination imbalance occurs.

[0050] Subsequently, based on the detected problem area, the backlight correction unit (200) approximates the observed values ​​of the image as the product of the illumination component and the reflection component for modeling, performs brightness restoration and illumination component estimation, and then preserves the reflection component to extract an illumination-invariant color map that is not affected by changes in illumination. At this time, normalization is performed based on the average illumination to effectively correct brightness distortion caused by backlight, polarization, or non-uniform illumination only in the problem area. Finally, the fusion unit (300) integrates the restored correction area and the non-problem area where the original is preserved to generate a final image, and by naturally combining them to minimize the sense of incongruity or artificial traces that may occur at the boundary of the two areas, an improved image can be output in which overall visibility is enhanced and the balance of color and brightness is maintained.

[0052] The region division unit (100) can detect a problem area where illuminance imbalance has occurred by analyzing the luminance distribution in pixel units from an input image frame. Specifically, the region division unit (100) is configured to precisely identify the area where illuminance imbalance has actually occurred within the image by analyzing the input image frame in pixel units, and can detect the problem area based on the statistical deviation of the local luminance distribution rather than a simple global brightness judgment.

[0053] In other words, by considering pixel-level luminance change patterns and spatial distributions rather than relying on the average brightness of the entire image or a single threshold, it is possible to distinguish between areas that are relatively dark due to backlighting and saturated areas that are excessively bright, thereby selecting only the areas that actually require correction. This configuration has the technical effect of preventing unnecessary image distortion and improving the selectivity and precision of the post-correction process, and can be clearly distinguished from conventional batch correction methods.

[0055] The region division unit (100) can calculate a luminance distribution by performing a histogram-based statistical analysis on the RGB channels of the input image and generate a luminance map based on pixel brightness based on the luminance distribution. Specifically, the region division unit (100) can quantitatively determine the distribution characteristics of pixel brightness within a single frame by performing a histogram-based statistical analysis on each RGB (Red, Green, Blue) channel of the input image. That is, by analyzing the brightness frequency distribution and cumulative distribution for each RGB channel without relying on a simple average value or a global threshold, and by precisely calculating the degree and location of luminance deviation within the image, it is possible to more reliably identify areas where brightness distortion has occurred due to backlighting or saturation.

[0056] Furthermore, by generating an illuminance map based on pixel brightness using these statistical analysis results, the illuminance distribution according to the spatial location of the image can be clearly represented visually and numerically, and the area to be corrected can be accurately defined in the subsequent correction step. This configuration improves the accuracy of determining illuminance imbalance and prevents unnecessary corrections, thereby enhancing the preservation of image information, and distinguishes it from conventional simple threshold-based segmentation methods.

[0058] The region division unit (100) can separate the brightness channel by converting the illuminance map into a color space, generate a backlight region mask and a saturated region mask respectively according to a preset threshold, and combine them to generate a final problem region mask. Specifically, the region division unit (100) can determine the problem region using only pure illuminance information, regardless of the influence of color components, by clearly separating the brightness channel through color space conversion based on the illuminance map. That is, unlike the conventional method of directly determining in the RGB space, by converting to a color space where brightness components are expressed independently and analyzing based on the brightness channel, it is possible to effectively reduce false detections caused by color deviation or the inherent color of the object.

[0059] Furthermore, by utilizing preset thresholds to generate independent masks for relatively dark backlit areas and saturated areas where detail is lost due to excessive brightness, and by computationally combining them to construct a final problem area mask, different types of illumination distortion can be comprehensively identified simultaneously. This multi-mask generation and combination structure enables more precise delimitation of the correction target area, thereby improving the selectivity and stability of post-correction and clearly distinguishing it from area determination methods based on a single condition.

[0061] The backlight correction unit (200) can mitigate brightness distortion caused by backlight or saturation by separating the lighting component and the reflection component based on the brightness channel in response to the problem area detected by the area division unit (100), and correcting the lighting component while preserving the reflection component. Specifically, the backlight correction unit (200) can separate the lighting component and the reflection component limited to the problem area detected by the area division unit (100), and selectively correct only the lighting component while preserving the reflection component. This is implemented by interpreting the observed value of the input image as a combined result of lighting and reflection, and correcting only the lighting component distorted by an external light source while maintaining the reflection component containing the subject's unique color and texture information. This configuration allows the relative color ratio and contour information of the subject to be stably maintained even in a backlight environment, and has the effect of suppressing excessive brightness amplification or color bleeding. In particular, since the correction operation is performed only on the problem area, the computational efficiency is high, and it is suitable for real-time processing environments.

[0063] The backlight correction unit (200) approximates the observed value of the input image as the product of the illumination component and the reflection component, and normalizes the illumination component to a reference value to generate an illumination-invariant color map, thereby maintaining the relative color ratio of the subject even in a backlight environment. Specifically, the backlight correction unit (200) models and interprets the observed value of the input image as the combined result of the illumination component and the reflection component, thereby structurally separating and processing brightness distortion caused by backlight. That is, by distinguishing between the illumination component, which fluctuates due to changes in the intensity or direction of an external light source, and the reflection component, which contains the subject's unique color and texture information, the problems of color distortion and loss of detail that occurred in a simple brightness amplification method can be effectively avoided.

[0064] Furthermore, by generating an illumination-invariant color map through normalization of lighting components to reference values, the relative color ratios between pixels are stably maintained even in backlit environments, thereby ensuring visual consistency with the subject's original color. This configuration enables color representation robust to changes in illumination, which can improve the reliability of image analysis and object recognition.

[0066] The illumination-invariant color map can be selectively applied only to the area where the problem area mask is applied, and the original image can be maintained in areas other than the problem area. Specifically, the backlight correction system (10) can be controlled so that the illumination-invariant color map is not applied uniformly to the entire image, but is selectively applied only to the area where the problem area mask is applied. That is, the illumination-invariant color correction is performed only on the area where actual visibility is reduced due to backlight or saturation, and the pixel values ​​and color information of the original image are maintained in the area where no illumination distortion occurs, thereby preventing unnecessary changes in image quality or over-correction.

[0067] This selective application structure significantly enhances image information preservation and structurally blocks issues such as color bleeding, contrast distortion, and artificial image alterations that may occur in normal areas. Furthermore, the control method, which clearly distinguishes between problem and non-problem areas and applies different processing paths, contributes to ensuring correction stability and consistency and can be clearly distinguished from global correction methods.

[0069] The backlight correction unit (200) can perform brightness restoration operations in a logarithmic space to enhance details in areas that are excessively dark from a set standard, and can balance brightness contrast by suppressing highlights in saturated areas through brightness tone mapping. Specifically, the backlight correction unit (200) can perform brightness restoration operations in a logarithmic space rather than a linear space. Operations in a logarithmic space can relatively amplify the response to low brightness areas and smoothly adjust the rate of change to high brightness areas, thereby effectively enhancing fine details in areas that have become excessively dark in a backlight environment.

[0070] In addition, by suppressing highlights in saturated areas through parallel brightness tone mapping, extreme contrast between bright and dark areas can be mitigated, and the overall image contrast balance can be stably adjusted. Unlike simple brightness amplification or contrast adjustment, this dual processing structure is effective in simultaneously achieving shadow restoration and highlight suppression, while improving visibility while minimizing image distortion.

[0072] The backlight correction unit (200) can achieve improved visibility even in environments with lighting changes exceeding a set standard by dynamically adjusting the correction intensity according to at least one of the frame-by-frame average illuminance change of the input image or the area ratio of the problem area. Specifically, unlike conventional methods that use fixed correction parameters, the backlight correction unit (200) can dynamically adjust the correction intensity by calculating the frame-by-frame average illuminance change of the input image or the area ratio of the problem area in real time.

[0073] In other words, the correction intensity automatically adapts even when the lighting environment changes rapidly or the backlight area expands or contracts, preventing excessive or insufficient correction. This dynamic control structure quantitatively reflects the degree of illumination change within the image and continuously adjusts the correction level, thereby enabling stable visibility improvement even under lighting conditions that change over time. Furthermore, through frame-by-frame adaptive control, flickering or discontinuous image quality changes can be suppressed, ensuring the visual consistency and reliability required for real-time surveillance video.

[0075] The fusion unit (300) can generate a final image by naturally combining the problem area corrected by the backlight correction unit (200) and the non-problem area of ​​the original image through boundary blending. Specifically, the fusion unit (300) can reduce visual discontinuities or artificial boundary phenomena that may occur at the boundary area, rather than simply combining the problem area corrected by the backlight correction unit (200) and the non-problem area of ​​the original image.

[0076] To achieve this, boundary blending is applied to ensure that changes in brightness, color, and contrast are spatially and gradually connected, thereby minimizing the disparity between the corrected and uncorrected areas. This fusion structure significantly enhances the naturalness of the image correction results and has the effect of increasing visual reliability in footage that is viewed for extended periods or used for analysis, such as surveillance footage. This boundary processing concept, which is distinct from simple synthesis methods, can provide clear technical benefits in terms of image quality.

[0078] The fusion unit (300) may apply at least one of weight-based blending or multi-resolution mixing techniques to reduce artificial boundary phenomena that may occur at the boundary between the corrected area and the uncorrected area. Specifically, the fusion unit (300) is not a structure that simply connects the corrected area and the uncorrected area, but can reduce artificial boundary phenomena such as brightness discontinuity, color breakage, and contour distortion that may occur at the boundary between the two areas.

[0079] To this end, when applying weight-based blending, position or distance-based weights are assigned to adjacent boundary regions to allow the corrected image and the original image to blend gradually, thereby minimizing visual dissonance. Additionally, when applying multi-resolution blending techniques, image information extracted from different spatial resolutions is combined stepwise to maintain a natural flow of light and shadow in low-frequency components while stably connecting boundary details in high-frequency components. This boundary-centered fusion structure provides the effect of suppressing artificial traces while maintaining the correction effect, and can be clearly distinguished from simple synthesis methods.

[0081] The control unit (400) is connected to the region division unit (100), the backlight correction unit (200), and the fusion unit (300), respectively, and can control the operation of the region division unit (100), the backlight correction unit (200), and the fusion unit (300). Specifically, the control unit (400) is a core component that integrally controls the operation of the region division unit (100), the backlight correction unit (200), and the fusion unit (300), and performs the role of dynamically adjusting the operation status and correction range of each component according to the existence and characteristics of the problem area within the image. In particular, the control unit (400) prevents the original image information of the non-problem area from being unnecessarily altered by controlling the backlight correction unit (200) to selectively perform correction only on the problem area.

[0082] This has the effect of simultaneously ensuring the information preservation and processing stability of the image, and can structurally resolve the overcorrection problem that occurred in existing automatic correction technology. In addition, this central control structure can improve the overall reliability of the system by maintaining consistent correction quality even in various image input environments. The control unit (400) is equipped with a wireless communication module so that it can be linked with external mobile terminals, etc., to control operation or check the control status in real time from the outside.

[0084] The backlight correction unit (200) can selectively perform correction only on the problem area under the control of the control unit (400) and maintain the original image information of the non-problem area.

[0086] The backlight correction system (10) is mounted on at least one of a CCTV camera (30), a network video recorder (NVR) (40), and an edge computing device (50) to perform backlight correction on an input video stream in real time. Specifically, the backlight correction system (10) is directly mounted on at least one of a CCTV camera (30), a network video recorder (NVR) (40), or an edge computing device (50) to perform backlight correction on an input video stream in real time.

[0087] Unlike offline post-processing or server-based batch processing methods, this minimizes latency by enabling immediate correction during the video generation and transmission stages, providing a structure suitable for real-time monitoring and analysis environments. Furthermore, it can be flexibly applied to various system layers, such as camera, storage, or edge device levels, ensuring high compatibility with existing infrastructure and maintaining stable correction quality even in distributed processing environments. This real-time stream processing and on-device integration structure can significantly enhance the practicality and scalability of backlight correction technology.

[0089] Referring to FIG. 3, as illustrated in the flowchart above, the image processing method according to the present invention first performs a step (S100) of collecting an original image through an input device or a shooting device, and includes a step (S200) of calculating an overall illuminance distribution by analyzing luminance values ​​in pixel units or area units for the collected image. Subsequently, based on the illuminance distribution analysis results, an image segmentation step (S300) is performed to divide the image into areas with relatively high luminance and areas with relatively low luminance, wherein the segmentation criteria may be determined according to a set threshold or statistical distribution characteristics.

[0090] Next, a correction step (S400) is performed for low-luminance areas to enhance visibility while minimizing the loss of dark area information, by performing brightness enhancement or contrast correction. Simultaneously, a step (S500) is performed for high-luminance areas to prevent saturation or contrast distortion caused by excessive brightness, by performing brightness compression or distortion suppression processing. Subsequently, a step (S600) is performed to normalize the brightness balance of the entire image by naturally fusing the low-luminance area correction results and the high-luminance area distortion suppression results at the boundaries, thereby minimizing artificial boundaries or dissonance caused by local correction. Finally, a step (S700) is performed to output a final image with balanced overall brightness and contrast, thereby providing an image with improved visibility even in backlit or illuminance-unbalanced environments.

[0092] As described above, the detailed description of the preferred embodiments of the present invention disclosed is provided to enable those skilled in the art to implement and practice the present invention. Although the present invention has been described with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the present invention without departing from the scope of the invention. For example, those skilled in the art may utilize each configuration described in the embodiments described above in combination with one another. Accordingly, the present invention is not intended to be limited to the embodiments shown herein, but to be given the broadest scope consistent with the principles and novel features disclosed herein.

[0093] The present invention may be embodied in other specific forms without departing from the spirit and essential features of the invention. Accordingly, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention shall be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention. The invention is not intended to be limited to the embodiments shown herein, but to be given the broadest possible scope consistent with the principles and novel features disclosed herein. Furthermore, embodiments may be constructed by combining claims that are not explicitly related in the claims, or by including them as new claims through amendments made after filing. Explanation of the symbols

[0094] 10: Backlight Correction System 30: CCTV camera 40 : Network Video Recorder (NVR) 50: Edge computing devices 100 : Area division part 200 : Backlight correction unit 300 : Fusion section 400 : Control unit

Claims

Claim 1 A backlight correction system for improving the visibility of an input image captured in a backlight environment, comprising: a region segmentation unit that analyzes the luminance distribution in pixel units from an input image frame to detect a problem area including a backlight area and a saturated area; a backlight correction unit that separates an illumination component and a reflection component based on a brightness channel in correspondence with the problem area detected by the region segmentation unit, and corrects the illumination component while maintaining the reflection component; and a fusion unit that generates a final image by applying boundary blending to the boundary area between the problem area corrected by the backlight correction unit and the uncorrected non-problem area of ​​the original image. A selective backlight correction system for improving image visibility, comprising: a control unit connected to the region division unit, the backlight correction unit, and the fusion unit, respectively, and controlling the operation of the region division unit, the backlight correction unit, and the fusion unit; wherein the backlight correction unit selectively performs correction only on the problem area under the control of the control unit, maintains pixel values ​​and color information of the original image for the uncorrected non-problem area, and dynamically adjusts the correction intensity according to the frame-by-frame average illumination change of the input image. Claim 2 A selective backlight correction system for improving image visibility according to claim 1, wherein the region division unit performs histogram-based statistical analysis on the RGB channels of an input image to calculate a luminance distribution and generates a luminance map based on pixel brightness based on the luminance distribution. Claim 3 A selective backlight correction system for improving image visibility according to claim 2, wherein the region segmentation unit separates the brightness channel by color space conversion of the illuminance map, generates a backlight region mask and a saturated region mask respectively according to a preset threshold value, and combines them to generate a final problem region mask. Claim 4 A selective backlight correction system for improving image visibility according to claim 1, wherein the backlight correction unit approximates the observed value of an input image as the product of an illumination component and a reflection component, and generates an illumination-invariant color map by normalizing the illumination component to a reference value, thereby maintaining the relative color ratio of the subject even in a backlight environment. Claim 5 A selective backlight correction system for improving image visibility according to claim 4, characterized in that the illumination-invariant color map is selectively applied only to the area where the final problem area mask is applied, and the original image is maintained in the area outside the problem area. Claim 6 A selective backlight correction system for improving image visibility according to claim 1, wherein the backlight correction unit performs a brightness restoration operation in a logarithmic space to enhance details in areas that are excessively dark from a set standard, and balances brightness contrast by suppressing highlights in saturated areas through brightness tone mapping. Claim 7 delete Claim 8 A selective backlight correction system for improving image visibility according to claim 1, wherein the fusion unit applies at least one of weight-based blending or multi-resolution mixing techniques to reduce artificial boundary phenomena that may occur at the boundary between a corrected area and an uncorrected area. Claim 9 A selective backlight correction system for improving image visibility according to claim 1, wherein the backlight correction system is mounted on at least one of a CCTV camera, a network video recorder (NVR), and an edge computing device, and performs backlight correction in real time on an input video stream.

Images